Quince Therapeutics, Inc. (QNCX) Earnings Call Transcript & Summary

Good morning, and welcome to Innovations in Alzheimer's Disease: Getting to know the root cause of Neurodegeneration hosted by Cortexyme. [Operator Instructions] As a reminder, this event is being recorded, and a replay will be made available on the Cortexyme website following the conclusion. I would now like to turn the call over to your host, Casey Lynch, Co-Founder and Chief Executive Officer of Cortexyme. Please go ahead, Casey.

Thank you, Sarah, and welcome to everyone joining us today. Next slide, please. I'm very pleased to be joined today by Dr. Marwan Sabbagh. Dr. Sabbagh is considered one of the leading experts in Alzheimer's and dementia, and is the Endowed Chair for Brain Health and Director of Translational Research at the Cleveland Clinic Lou Ruvo Center for Brain Health in Las Vegas. Dr. Sabbagh has dedicated his career to finding a cure for Alzheimer's and other age-related degenerative diseases and hopes one day to work himself out of a job. Dr. Sabbagh is a leading investigator for many prominent national Alzheimer's prevention and treatment trials. He's co-authored and authored almost 400 medical and scientific articles on Alzheimer's research. Also joining us today is Cortexyme's Chief Medical Officer, Dr. Michael Detke. Mike previously spent a decade at Eli Lilly, leading clinical development of Prozac and Cymbalta, prior to becoming CMO at multiple smaller biotech companies, including in the Alzheimer's arena. You can see our agenda on the right slide -- side of the slide. I'll give a very brief overview of the company, followed by remarks from Dr. Sabbagh about the current environment in Alzheimer's disease drug development, presentation of previous and new data we released this week at AAIC by Dr. Detke, and a preview of the GAIN trial baseline data and important aspects of the trial designed by Dr. Sabbagh, followed by time for questions. Next slide, please. I'll be making some -- we may be making some forward-looking statements today. Next slide. As most of you know, we're looking forward to top line data in a few short months for atuzaginstat in mild to moderate Alzheimer's disease during Q4. This is a well-powered 643 subjects pivotal efficacy study using cognition and function as the co-primary endpoints. The company is focused on upstream mechanisms of action. As you'll hear more about today, we believe that this -- the bacteria P. gingivalis that we have discovered in the brain of Alzheimer's patients is driving downstream pathology, and we'll present that evidence today. The work the company is doing is all based on our own proprietary small molecules. These are novel small molecules that are brain penetrant. There's no requirement for IV infusions or MRI monitoring. There's an expanding evidence base around our approach from our own research and third-party literature. We also have a growing pipeline. We'll also be announcing top line data from a substudy, a Phase II substudy in periodontal disease called the REPAIR trial in Q4, and we're starting up a Phase II in Parkinson's disease with the Parkinson's study group. Finally, we recently announced selection of a lead molecule for coronavirus treatment, which we'll be progressing. We're well funded to execute on all of this work with $170 million in cash. Next slide, please. I'll now turn it over to Dr. Sabbagh.

Thank you, Casey. Next slide, please. So we have lived in the era of symptomatic drugs. This is the cholinesterase Inhibitor/memantine for over 2 decades. And of course, recently, we had the approval of aducanumab ADUHELM as a monoclonal antibody to reduce the rate of decline by removing amyloid. As you see in this slide, of course, there is some -- despite all the firestorm controversy, there's also safety concerns that we have to consider. And so we -- I don't know that this field has settled on, this single monoclonal. I do want to disclose that I have a relationship with Biogen and will disclose if other relationships come up. But the point is, is that there are relatively few drugs approved, one targeting amyloid, the rest of them are symptomatic drugs, which leaves the Alzheimer's disease still, despite June 7, as a huge area of unmet need. Next slide. Every year, Dr. Cummings and our teams publish the Wheel of Fortune, as I call it, in the Alzheimer's space. This is the 2020 iteration. But we talk about monoclonal antibodies, small molecules. The outer ring is Phase I, middle ring is Phase II and inner ring is Phase III, whether it's a biologic, symptomatic or small molecule targeting. The point is that this is a very active area of research. There's a lot of drugs in development, and you see very clearly that COR388 is right in the middle of the pack in the Phase III space and is worthy of consideration. Next slide. So this is a very important concept. We always said plaques and tangles since the lowest Alzheimer's. But what we know now, of course, is that you -- amyloid is a much more complex biology. The amyloid accumulation may trigger astrocyte activation, which will be in response to things like tau as well and microglial activation, neural damage, et cetera. But it turns out that P. gingivalis and gingipains may be working the upstream. Some people have posited the possibility that amyloid itself has antimicrobial aspects, and the question then becomes is it gingipains and P. gingivalis one of the inciting events that could lead to downstream pathology, and that makes this very exciting because that would suggest that COR388 has the potential to have disease slowing properties by hitting multiple targets and -- or at least slowing the cascade on multiple targets and having effects on multiple pathogenic mechanisms. Next slide. There is extensive evidence from several scientific domains supporting the gingipain hypothesis. Important thing you need to understand in this audience is that a lot of this literature is not published or generated by Cortexyme. Most of it is not generated by Cortexyme, which means it's independently established and extensive third-party publications based on extending and replicating the Cortexyme data really enforces the legitimacy of their findings. And specific examples, of course, as we just said, is that Alzheimer's disease has itself an amyloid elements of infection or residua to infection, including microglia activation, inflammasome, complement, that it might be itself an antimicrobial peptide. I don't think everybody agrees or has come to consensus on that, but that has been proposed by several prominent laboratories in the country, and that supports that infection triggers could be a cause of upstream mechanism, that periodontal disease, which is caused by P. gingivalis is a strong predictor of AD. And some of the data that Mike Detke is going to show you will reinforce that, and that it is seen in almost every brain of Alzheimer's patients, which would suggest this is not a coincidental finding and might be mechanistically overlap. And I think they could -- that Cortexyme can now make that argument very convincingly, and multiple other pathways need to be considered. So what I'm trying to say is that P. gingivalis and gingipains are not a spurious off finding, but are core to the pathophysiology of Alzheimer's disease. Next slide. So there is an area -- a huge area of unmet need. As I mentioned before, the drugs have limitations, and it's important to consider that several things could affect downstream pathologies, and COR388 supports the gingipain hypothesis and can block gingipains as a neurotoxin. And I think Mike will talk a little bit about that. So gingipain is a protein that's coming out of P. gingivalis, which has neurotoxic properties, and it might be a very good target to pursue, and this is what makes this so exciting. Next slide. And with that, I turn it over to Mike Detke.

Thank you, Marwan. Next slide, please. As Dr. Sabbagh was saying, there are several different lines of evidence that support the gingipain hypothesis, and one of the first that was identified by our Co-Founder, Steve Dominy, is that periodontal disease is highly correlated with Alzheimer's disease, along with low educational level. It is one of the 2 strongest predictors in the population of Alzheimer's. And the publications on the left side of the screen are a few of the several that have been published and point this out. And while they're primarily correlational, some of them, like the third one down, showed that periodontal disease happens in early mid-adulthood and predicts poor cognitive function in men as they age further. And on the right are a few of the more recent publications showing that A-beta may function as an antimicrobial peptide. And as Dr. Sabbagh pointed out, all of these publications are by third-party academic KOLs, rather than Cortexyme. Next slide, please. And Dr. Sabbagh foreshadowed this nicely, too. If you focus on the right side of this slide to begin with, you can see that we know there are tau and amyloid in the Alzheimer's patient. However, there are a whole lot of other things going on that have not historically been well explained by existing theories, including fragmentation of ApoE and lysosomal dysfunction in neurodegeneration, but also a whole number of hosts or human or immune system responses. The host response is listed here including microglial activation, multiple markers of neuroinflammation, complement and inflammasome activation and so forth. And all of these are actually really nicely explained by an upstream infection portrayed on the left here with P. gingivalis. Occam's razor dictates that the simplest or most parsimonious explanation that explains the most data is all of the things being equal, the best one. And as you can see here, this infection explains a whole array of phenomena in Alzheimer's disease that have been troubling the field for some time. Next slide, please. Now it's very important to understand the mechanism of action of P. gingivalis and gingipain, and very interesting science, too. So if you cast your attention over to the left side of the screen here and the lower or intracellular side of this cartoon, you can see P. gingivalis bacteria are portrayed as these pink sort of pill-shaped structures here, and they have 2 very unusual features for a bacterium. One is that, as you can see here, it gets inside of cells. It's an intracellular bacterium. And secondly, P. gingivalis is asacrolytic, which means it does not live off of sugars or carbohydrates for its fuel source, it lives off of proteins. How does it do that? It releases these gingipains, which are proteases that are portrayed as orange little Pacman here because like proteases do, they go around chopping up proteins on the inside of cell. So because of P. gingivalis' nature, it's releasing these gingipain proteases that are chopping up proteins and really digesting the cell from within. I sometimes use the metaphor that this is like having termites in your brain cells or in your gum tissues. And it's been known for a long time in the dental literature that you really can't eradicate this bacterium. You would think antibiotics would do it, but it turns out they really don't, and we know this because antibiotics don't cure periodontal disease, and there are several reasons for that. One is that the bacterium is intracellular, where antibiotics can't always reach. It lives in the mouth in plaques and biofilms that, again, antibiotics can't reach. It can become dormant and develop antibiotic resistance. So instead of targeting P. gingivalis, Cortexyme developed drugs, like COR388 or atuzaginstat, on the right, portrayed as these little diamond gem-type features that bind covalently with the gingipain proteases. And this -- so this has 2 effects. Because the gingipain proteases are causing all the damage, once you deactivate them, all the damage is stopped. The other is that because they're chopping of proteins to provide food to the bacterium, they starve off the food supply to the bacterium, so you'll see that the bacterium level gets knocked down. Next slide, please. So we went looking for gingipains and P. gingivalis in the brains of Alzheimer's patients, and this is a nice portrayal visually of some immunohistochemistry for gingipains. And the immunohistochemistry causes the gingipains themselves to stain brown. So you can see in gingival infection on the far left frame, the brown staining in gum tissue, gingival tissue. And in the center frame, you can see the exact same brown staining with gingipains in the cells, in the hippocampus, the key memory area of Alzheimer's patients. And on the right is a control hippocampus, and you can see no sign of gingipains there. Next slide, please. But we wanted to look at more Alzheimer's brains and confirm or disconfirm this hypothesis that gingipains were in the brains of Alzheimer's patients specifically, so we collaborated again. This is an outside collaboration with the University of Auckland and New Zealand. They have a brain bank and allowed us to -- or use our immunohistochemistry for gingipains to interrogate this brain bank. And if you focus on the figure on the left first, and the fact that there's a logarithmic scale on the y-axis, essentially, below 1 is background noise, and above 1 means that the patient is infected with Pg and gingipains. So you can see that the Alzheimer's group here, virtually all of them are infected, about 95%. And you can see that it's highly statistically significant compared to age-matched controls with a P of 0.0001 down there at the bottom. But very importantly, you can see that about roughly 30% or 1/3 or so of the age-matched controls are above 1, meaning they are infected with Pg and gingipain, and that's exactly consistent with the upstream infection hypothesis that we've been showing you. We know that tau and amyloid and other signs of pathology in Alzheimer's can precede the cognitive symptoms by years or even a decade or 2. So you would expect, if the infection were upstream that there would be patients who had the infection and then died from some other cause in the years before they would have developed Alzheimer's and donated their brains to the brain bank. And importantly, also on the right slide, you can see that the gingipain load correlates with the load of tau and other biomarkers, like ubiquitin, that, in turn, correlate with clinical symptomatology. So these are all important supportive findings. Next slide, please. And then this is a mouse model, and Cortexyme has done this mouse model. And again, importantly, this has been replicated by 5 independent outside labs by now. And this is -- the data we're showing you right now is actually from the University of Illinois at Chicago on external lab. And this mouse model is simple and elegant, but powerful. The mouse model simply takes wild-type mice, as you can see up on the title, and rubs P. gingivalis on their teeth and guns. Nothing else. It's not put into the brain, just in the mouth. But within weeks afterwards, you can see that P. gingivalis infiltrates the brain, causes neuroinflammation by several biomarkers, including here is TNF alpha. It causes the uptick of A-beta plaques. It activates microglia, the immune cells of the brain. It causes tau tangle-like neurons with classic silver staining. And perhaps, most importantly, it causes the death of neurons in the hippocampus, again, key memory area, the CA1 and dentate gyrus region of the hippocampus. This is really unprecedented. Previous mouse models have shown increase in amyloid and maybe 1 or 2 other things, maybe reduced cognitive function. But here, we -- this very simple oral infection recreates all of the pathophysiology of Alzheimer's, again, explaining most of the data as per Occam's razor. Next slide, please. And now here's the same mouse model. However, we treated for a shorter period this time, and these are internal Cortexyme data. So you can see that the infection was for 6 weeks and then we treated with COR388, or atuzaginstat, for 5 weeks. So some of the differences here are a little bit smaller because of the shorter duration, but the important thing is that, while the -- on the far left panel, while the infected mice, you see significant increase in -- and presence of P. gingivalis in the brain, after they're treated with COR388, you see that the -- because the food supply has started off, the bacteria load goes way down. And again, the increase in amyloid beta, and then it's reversed by treatment with 388. The increase in inflammatory markers, like TNF alpha, and then it's reversed by 388. The reduction or death of hippocampal interneurons, which are then protected by 388. So across the board, you see a reversal of all of this pathophysiology of Alzheimer's by COR388. Next slide, please. And we continue, as more data come out in the literature, as good scientists, we're trying to find out whether this -- again, whether all these data are explained by the -- and consistent with the gingipain hypothesis. So last year, there were some really important publications showing that pTau 217 is an important new biomarker to pay attention to. And as you can see in the data here from last year that on the left, pT217 predicts the Alzheimer's disease as compared to controls, and on the right that it correlates with Braak stage, so an important validated biomarker. And in the next slide, so we went to say, "Well, does -- do gingipains seem to be related to this and cause this?" So we, again, infected mice and, again, treated with COR388. And you can see in the middle panel here that infection with P. gingivalis causes an increase -- significant increase in pTau 217. And then treatment with COR388, again, reverses it nicely. And on the right, it's just a replication where we're looking at the pTau 217 to total tau ratio. These were the -- some of the new data we presented at AAIC this past week. Next slide, please. Now turning to some of the early human data we collected. Again, consistent with this hypothesis, what we see here, this is in our Phase Ib multiple ascending dose study with Alzheimer's patients, the Alzheimer's patient cohort, and on the left is portrayed RANTES, also known as CCL5, which is an inflammatory biomarker. Therefore, you would predict that treatment with COR388, treating the infection that's upstream would reduce inflammation in these Alzheimer's patients, and that's exactly what we saw. We saw about a 30% reduction in 28 days of treatment, statistically significant reduction in RANTES. And on the right, CSF ApoE fragments, we know that P. gingivalis gingipains fragment ApoE. And so we know -- we hypothesized that in the study, by treating with COR388, we would reduce the number of fragments. And again, that's exactly what we saw, again, reducing by about 30% in this case in 28 days and, again, statistically significant drug versus placebo. Next slide. We've gone on to show that not only do gingipains fragment ApoE, but they preferentially fragment ApoE4 over ApoE3. You can see the -- these are -- this is mass spectrometry detection of these peptides, and the E30 is after 0 digestion and E31 is after 1 minute, E4 and E41 is ApoE4. So you can see that the gingipains digest or fragment ApoE4 preferentially to ApoE3, and this is because they cleave at lysine and arginine, and they have more lysine and arginine cleavage opportunities, in part because of the folding of ApoE leaves more of those on the surface. So this is an important explanation of the largest genetic risk factor for sporadic Alzheimer's, which is about 99% of Alzheimer's that's been out there for years, the ApoE4 risk factor. To our knowledge, this is one of the most, again, parsimonious explanation of this well-known important genetic risk factor. Next slide, please. And then turning to the clinical data with the same early phase Alzheimer's cohort on the left is the MMSE, and you can see a nice trend toward improvement in the treated group. The baseline to endpoint change in the treated group was almost significant here at a p of 0.052, but the drug placebo difference was trending nicely, but not significant. The CANTAB Memory Composite, again, the treated group in pink performed numerically better than the placebo group. And on the far right is a company called Winterlight has developed an AI-driven speech analysis program that categorizes several different components of speech. And treatment with COR388 was statistically significantly superior to placebo in this group in 3 different measures of this, 2 of them were still significant after a rather conservative bond for only correction. And this is one here. And importantly, too, the speech that they saw that was improved, this is prepositions and conjunctions, which correlates pretty well with sense and structure and complexity, which you see go downhill as Alzheimer's patients progress. And the other one was similar to -- was correlated with word finding difficulty, which also is an early sign of progression in Alzheimer's. Next slide, please. And this is an actual example of how the Winterlight path works. It's very simple. The patient has shown pictures like these and then ask to describe them as much as possible in a few minutes. And these are actual transcripts of a 75-year-old woman with mild to moderate AD in this study. And at day 1 prior to treatment, you can see she's saying the ellipsis here are pauses, so she pauses, chair, pause, plant, pause, couch. So you can see that the pauses indicate that word finding difficulty. And you can see there's virtually no sentence structure here. She's speaking in single words. Whereas after 28 days of treatment, you can see much more easier word finding, fewer pauses and more complexity in sentence structure, a rowboat with 2 people about to fall off of a ramp. And the bolded words are prepositions as per the data in the last slide. Next slide, please. And with that, this has all led us up to the now ongoing GAIN trial, and I'll turn it back over to Dr. Sabbagh to tell you about the design of that trial and some of the early data.

Thank you, Dr. Detke. So I'm going to walk you through some of the trial design elements, next slide, of the Cortexyme GAIN trial. This is a Phase II/III study. I want to just spend a few -- a minute talking about this slide because it's important is that on the right-hand side, you see the primary endpoints as ADAS and ADCS-ADL. The one thing I like about this design is people won't have to spend a lot of time trying to interpret it. This is a -- considered to be the gold standard. It can be clearly interpreted, and that's important so that we won't have to be interpreting composite scores or exotic statistical analyses. Second thing is it's a 3-arm study, it has a low dose, high dose and placebo with a mini-mental range of 12 to 24. And this is critically important because people are -- I think the monoclonal story has dominated the field of late, but this population is not overlapping with the monoclonal population. So we're clearly identifying a different population here, and that's an important because, again, it's addressing everybody's unmet need. The third thing is there's a substudy which shows that there is periodontal disease in 233 patients. There were -- traditional biomarkers are being accrued, but they're also looking at biomarkers of P. gingivalis infection. This has both Phase III and a Phase II element of dose finding. And the important thing about this, of course, is the -- this was powered to have 573, but they overenrolled kind of with the not to COVID, and we're -- and they're expecting to report out data in the next quarter. Next slide. So let's talk about the demographics of this population. And I think this is critically important because, again, not only do you want to look at a design on the basis of the inclusion/exclusion criteria and the design on the basis of their instruments. You want to look at the population you've recruited. So if you're going on the left-hand side, first thing you notice is the average age is around almost 70. That's a typical Alzheimer clinical trial. Second is there are more women than men, that it would be appropriate for an Alzheimer clinical trial. Third, I think one of the strengths here, again, on the left-hand side is that their population is a little bit more diverse than we've seen in other clinical trials. That's a strength because, of course, one of the biggest criticism and as you all know DIE issues have become major issues in clinical research. This -- the GAIN study has addressed that organically. The third -- fourth thing I want to point out to you is that it's an evenly distributed study between mild and moderate Alzheimer's disease. And then the last thing I want to -- 2 other things I want to point out to is that the ApoE distribution is very, very typical, meaning roughly 2/3 are ApoE4 carriers and 1/3 are not. And the memantine and cholinesterase inhibitor use would be typical for a study. So what I'm trying to say to you is that the approach, the methodology, design and baseline characteristics do all represent what would be considered a standard Alzheimer clinical trial. Next slide. So natural question that has come up and continues to come up in studies similar to this is, did you select patients on the basis of biomarker inclusivity, as some studies -- as the field has moved to -- in a certain way? And the answer is no, not in a prespecified manner. But when you drill down and look at CSF on these patients -- on these participants, you see that 84% of the sample actually meets the CSF criteria for Alzheimer's disease. So in other words, using rigorous criteria, they have effectively picked the sample that they wanted and biologically that the overwhelming majority of the population in the study -- in the cohort actually would meet biological criteria for Alzheimer's disease. And this is critically important, so that you know you're selecting your -- the correct population. And I will say to you that their SAP, their statistical analysis plan has built in to look at it in both ways. Next slide. So another key element, and I believe you reported this out at this conference, Dr. Detke, is that 100% of patients enrolled in the subjects -- subjects enrolled in this study had P. gingivalis, most infected at levels associated with significant periodontal disease, which means that this study, the GAIN study is -- has a well-selected population to test the gingipain hypothesis. So not only this is a drug trial. This is testing a hypothesis about the link between gingipains and Alzheimer's disease. While people mounted an immune response to P. gingivalis infection, and this has been shown in the literature to correlate with the level of infection with the bacteria, has many immune evasion strategies, including expression of gingipains, which we know are neurotoxins, and it is known to persist even in the face of antibodies. Next slide. So this is the substudy on the periodontal disease. And you -- interestingly enough, what you see here is that almost every single person in the mild, moderate to severe Alzheimer's range actually have periodontal disease. So mechanistically, we're trying to reconcile the 2, but the -- you -- when you see the overwhelming data, you suggest that this is not an incidental or coincidental finding, but there is a mechanistic overlap, and these data support that observation. Next slide. So the GAIN trial leverages the gold standard for -- designed for a clinical trial. It's a well-accepted cognitive and functional co-primary endpoints. It has appropriate biomarkers. It has the correct population role with demographics and baseline biomarkers. And so it will be easy to interpret as a positive result. Hopefully, it is. Next slide. And this allows promise for progress, right? So the mechanism of action has potential to impact the A-beta tau, inflammation, lysosomal dysfunction, et cetera. It's a twice-a-day oral medication, which confers huge advantages over the drugs that are being developed, the monoclonals. There's no need for PET imaging. There's no need for safety monitoring using PET to look for ARIA, and the mechanism of action provides hope for a greater disease impact. And with, that slide, we can -- I will leave it to Casey to run us through some of the conclusions here, please.

Thank you, Marwan. So in conclusion, evidence continues to accumulate, supporting the gingipain hypothesis. The GAIN trial is designed to be a well-powered pivotal study with a very clear readout. Baseline data from GAIN are supporting that we've enrolled the right population at the right time to test atuzaginstat for Alzheimer's. The demographics are appropriate. The biomarkers are appropriate. The P. gingivalis infection load is really supportive of the hypothesis as well as the right population in the study. So we're really excited to be looking forward to data in just a few short months during Q4. And now we will take questions. Next slide.

Thanks, everyone. I think as everyone knows, you can use the web interface to punch in for questions, and I will ask and answer them if you'd like to remain anonymous. Just please let me know that. The first one comes from Jason Butler from JMP for Dr. Sabbagh, asking about your thoughts on the biomarker data that Cortexyme has shown. For example, the CSF A-beta reduction, ApoE fragmentation and phosphorylated tau. I'm going to let you -- let me keep going with that. But are there certain biomarkers that are more compelling in the context of the accelerated approval of ADUHELM based on A-beta as a biomarker?

Yes. So that's a complex, very well-informed question that -- but the answer is phosphorylated tau would be the one, and I would love Dr. Detke's perspective on that. But simply put, if you can reduce tau, I think that means you have target engagement and you're reducing downstream pathology. And so I keep an eye on phosphorylated tau. I would love Dr. Detke's perspective.

No, Dr. Sabbagh, I agree completely. As you know, tau and pTau and other biomarkers, like hippocampal shrinkage, which we're also monitoring in the GAIN trial are nicely correlated with progression of AD and level of symptomatology. And for those reasons -- which amyloid is not as closely correlated. So for those reasons, I agree with you completely. Thanks.

Next question from Dave Sherman at LifeSci Capital. Can you talk about some of the strengths of targeting the lysine gingipain as a key virulence factor?

Mike, do you want to take that question?

Sure. Yes, there -- we didn't go into too much detail about this, but P. gingivalis releases 2 kinds of gingipains, 1 cleaves at lysine and 1 cleaves at arginine. And you might initially think you need to stop both of these, but they are both released as pro-proteins, and they cleave each other to activate each other. So it turns out that both are necessary for the pathology, and neither is sufficient, so that knocking down the level of one of them or even eradicating is sufficient to really produce maximal efficacy. And we've seen this across multiple animal models and in vitro models that we've done with both the lysine and arginine gingipain inhibitors. So great question, but I'll stop there.

Yes. And I'll just add, this was published in detail in Dominy et al Science Advances paper, where we did some head-to-head combination, comparisons, including arginine versus lysine gingipain inhibitors and antibiotics as well. And lysine gingipain inhibitors were, by far and away, were very effective on that alone.

This one is probably for Dr. Sabbagh, but, Mike, you can also weigh in. It's asking about the Winterlight assessment, and could the improvement be a function of simply becoming familiar with the test? It could probably apply to some of the other Alzheimer's tests as well. But patients at 28 days have seen this originally unfamiliar test, but know what's expected going into the subsequent tests.

Yes. Great question, and that's a person who obviously follows neuropsych literature well. If you could actually pull up Slide 25, somebody? You're just -- right. here we go. Okay. So the answer to the question is, this is -- first of all, we use normative means on this. And second of all, if you notice, the images are different. So each of the -- each image actually has an estimated what would be a kind of suitable scoring, but the point is that you can measure complexities. And so there is no practice effect here, and that's an important question because if you give the same test over and over again. And I know this every day in my practice, when I hear my patients whispering to each other, ball, flag, tree, and apple, table, penny, they're practicing for taking the mini mental. So this is -- this particular instrument is appealing because it does not have embedded possibility of practice effect.

And I'd just add in that this is pretty standard in neuropsych testing that there can be practice effect. So you mitigate that by using different pictures at the beginning and end here. You counterbalance the picture, so some patients get this one at the beginning, that one in the end. And other patients get it in the reverse order. And the key comparison was really drug versus placebo, where we saw very statistically significant differences on very relevant measures specifically that have been correlated with Alzheimer's progression.

Our next question comes from Sumant Kulkarni from Canaccord. Dr. Sabbagh, you've been involved in a lot of clinical trials. This one is different in that the hypothesis involved gingipains and potentially has an effect on gum disease. So in your experience, what are the chances of functionally unblinding during the trial because patient's underlying gum disease may go away?

It's a critical question, and I believe there are mechanisms embedded in the trial to ensure that, that does not occur. But that is -- we always worry about studies like with the potential for unblinding. But I think that they have done a good job in trying to overcome this. And particularly, I think there -- the subcontracts, the investigators, the physicians, neurologists themselves or the clinical investigators are not the ones doing the oral assessment. Did you want to add anything to that, Dr. Detke?

No. That's a good point. The oral assessors are different sites than the Alzheimer's assessors. The only other thing I'd add is any drug, if you see efficacy, could theoretically be unblinded. So that's a problem with any -- in a theoretical problem with any study. But I'd also add that a really great finding in the periodontal literature would be changing gum progression by about 1 millimeter. So it's hard to imagine something that would be massively unblinding here.

And I want to also point out that from a neurologist standpoint, when I'm doing a clinical trial like the GAIN study, I will tell you that we're walking in, making sure there's no safety issues. We are not the ones assessing the gum and the periodontal disease. So -- and I'm not sure I would know what I'm looking for as a neurologist. But similarly, I'm not sure a dentists would know what to look for in an Alzheimer patient. So it all works out.

Next is just is a follow-up from Sumant of Canaccord. How can tools like Winterlight be adopted to non-English-speaking population in national trials? Maybe, Mike, do you want to take that one?

Yes. They can be translated and normalized, and I know that Winterlight is working on that in our trial to keep it simple and start at the beginning. We only did the Winterlight analysis in English-speaking patients, so that's a large subset of the patients in the United States and some in the United Kingdom.

A couple of questions here on safety. In general, please discuss the safety profile. Any potential ongoing safety concerns that you'd care to bring up that we are looking for a priority? I don't know if we want to start with Mike and then have Dr. Sabbagh's thoughts on that.

Sure. On a high level, the safety profile has been really outstanding. If you think about it, the mechanism, remember, this mechanism is targeting a protease that is unique to bacteria, and so it really doesn't target -- atuzaginstat doesn't target anything in the human body. We've assessed it against over 800 anti-targets, and there's at least 1,000-fold selected. The highest doses tested in the chronic mouse and dog toxicology tests were the NOAEL doses. So overall, the profile, we're extremely happy about. We have a press release and described the fact that we have had a few adverse events related to the liver. Given that this is still an ongoing blinded trial, we really can't talk about that in any more detail, but we're confident that it's manageable in the real world with things as simple as a simple blood test and titrating up the dose. Anything you wanted to add to that, Dr. Sabbagh?

No.

Dr. Sabbagh, you said at one point during your prepared remarks that the SAP has it built in to look at it both ways. What did you mean by that? And could you elaborate a little bit about that?

So -- and you can look at it by people who are biomarker. The sample, you can look at the sample, the analysis plan for the entire sample or just for the people who are amyloid, had CSF biomarker positive. So the SAP builds both into it. But I would say to you that the majority -- I mean, you're talking about almost 90% of the sample has Alzheimer pathology. So people will always question, well, is that 15% driving any of the clinical outcome or the interpretation? The answer for the most part is probably not.

Next question. On autopsy, do we know for sure that there is the presence of this bacteria in brain plaques of Alzheimer's patients? Casey?

Yes, I can take that. As Mike walked through the autopsy data and, again, published in detail in Science Advances, we looked at multiple different antigens, different proteins expressed by this bacteria and DNA from the bacteria. So yes, we're quite confident. It is in the brain of patients. We see it inside the neurons and inside astrocytes. So actually, it co-localizes with A-beta 42 inside the neurons. And we think, again, the neurons are creating the A-beta 42 as part of the antimicrobial response, whereas the plaques are external, actually not in contact with the bacteria in our studies.

A follow-up from the same individual asking about whether or not the people get better? Or are they just not getting worse? Mike, if you want to maybe start with that.

Sure. Our expectation is that by stopping the upstream infection, we should stop or greatly slow progression. And the GAIN trial is powered at 90% statistically to have a 90% chance to see a slowing of approximately 50%. However, it's mechanistically and regarding our Phase I data, the -- there's reason to think that you might see a little bit of improvement. The -- there are some neurons that are dead in the disease, but there are some that are heavily infected with Pg and chock-full of inflammatory response. And clearing up that inflammation could potentially bring those neurons back online. And that's consistent with the data that we saw in the Phase Ib data that showed in the direction of improvement in all 3 of the clinical outcomes we showed, in the mini-mental state examination and the CANTAB and the Winterlight speech analysis.

Next question comes from Neena Bitritto-Garg from Citi asking, though it's hard to predict what the FDA would be willing to do, do you think that they would accept a filing based on an analysis in just the biomarker positive patients? Or is success still defined as the entire population from a regulatory perspective?

Do you want me to start on this, Casey, and jump in?

That would be great.

Our study is designed with the clinical co-primaries of cognitive scale and a functional scale, which is the gold standard for assessing Alzheimer's. And should it be positive on that, clearly, that would be, we believe, considered a positive trial. So the regulatory landscape seems to have changed a lot in the last couple of months and allowing for the possibility that we could see improvement in a biomarker, too, like CSF A-beta, and that could be acceptable, too. So we're cautiously optimistic that with a great efficacy signal on either of those things that it could be considered potentially for a single trial.

A follow-up question from Dave Sherman at LifeSci Capital. Can you please talk about the 12-month time point with respect to the specific mechanism that we're discussing here? Why do we think this is enough time to dampen inflammation and for the placebo group to decline? Dr. Sabbagh, maybe you can start with that one, expected decline rates in placebo.

Yes. So the data suggests that the biggest predictor of progression is your baseline cognitive scoring. So the fact that they picked a mild to moderate, but skewing into the moderate range, really underscores the high probability of a group that will progress in the 12-month period of time. I think that's a great, great thing because you want to select a population that you know is going to progress. That way, you can test your hypothesis. The one frustration we've had, time and time again, but again in the mild stages is that you're picking a patient population that is not going to get worse no matter what. And so that's a very frustrating thing because then you're losing power. So in this case, of course, we're saying that I think you have predictive populations who are likely to progress. And I think the time frame of 12 months is a good estimate of time. And if the biological mechanism of reducing inflammation and infection actually and blocking the gingipain toxicity, I think 12 months should be sufficient to achieve that. I don't know, Mike, if you want to add anything to that.

No. That's perfect. I agree completely.

There's another question from Neena Bitritto-Garg at Citi on that topic and asking for Dr. Sabbagh's opinion, but probably Mike as well, what your thoughts on the rationale for not screening for baseline biomarker levels as entry criteria into the study and given the proposed mechanism, how will that impact the probability of success?

I want to comment on that. First, what we know is that people can have the P. gingivalis infection, regardless if they have amyloid or not. Everybody -- that's a common top conversation piece in the field is that you are -- are you selecting the right patient population. I want to comment, though, that we know that given the fact that you have the overall probability of the sample, having 85% of the sample having Alzheimer's pathology, you have picked the right population. And actually, if you look at the literature, only 15% seems to be appropriate for people with who would have other kinds of progressive conditions. So I understand the selection on a priority, but what I'm saying to you is that, regardless of -- or not selecting a priority, they have selected the correct population. So I think that the design has mitigated that concern, number one. Number two is the mechanistic suggestion is that people are going to have gingipain infection no matter -- P. gingivalis infection, no matter if they have amyloid pathology or not. Mike, did you want to add something to that?

No. That's good. I think if you have an amyloid-lowering drug, and that's your mechanism, you need to start with patients who have amyloid in their brain, regardless of whether that is correlating well with symptoms or not. We have Pg drug. All of our patients are infected with Pg, and 90% of them have moderate to severe periodontal disease, which is always caused by Pg. So I agree with everything you said, too, Dr. Sabbagh. Thanks.

This question follows from that somewhat. Assuming the GAIN trial is successful, and the drug is approved for someone of advanced stage who has gingivitis but not been diagnosed with AD, would you envision that atuzaginstat would be used as a prophylactic for Alzheimer's? Dr. Sabbagh, if you were...

So I hate speculating. I -- clinical trials and Alzheimer's is my day job. What I say is prove the concept here in the symptomatic population before we go chasing whether we should be looking at this as a prevention strategy. We envision these kinds of things down the road. But if you don't have a proof of principle in your target population, it's going to be hard to posit that it's going to work as a prevention strategy. So I'd say the answer is theoretically, yes. But in practice, let's get this first around baked.

I have a couple more questions and about 5 minutes left. This is a follow-up from Jason Butler at JMP for probably Casey. What are your plans for a companion diagnostic to identify patients with Pg infections and gingipains before they present with AD symptoms?

Great question. There's a lot of ways to diagnose Pg infection. Saliva test, our current favorite, is the Pg IgG as was shown in the presentation, antibodies against P. gingivalis indicating a systemic infection. We also have some exploratory cerebrospinal fluid test as well as we're working on a potential PET imaging agent, which would diagnose gingipains in the brain. So there's a lot of potential to -- for diagnosis, but what we're really learning from the GAIN trial, which is what we wanted to learn, is which is the best of those. And what we found is that a typical clinical diagnosis of AD -- a typical differential diagnosis based on the symptoms and exclusion of other things is really identifying these people who are infected with Pg. The coincidence of the biomarker with the symptomatology is so high that, probably, we don't need a companion diagnostic, although those will be available as well as if clinicians would like to use it.

All right. This will be the last question, unless someone punches in for a last minute one right now. But asking about the next steps after we get the results from the GAIN trial in Q4 in terms of FDA approval. And is the plan to eventually partner with a big pharma player? Or would you choose to bring atuzaginstat to market on your own? Casey.

Yes. We'll be engaging with the FDA immediately after the data, of course, for the fastest path to market. We're very happy to commercialize atuzaginstat on our own. That said, in light of positive data, there will likely be opportunities presented, and we will consider them as they come for the benefit of shareholders and the fastest launch of the drug.

Great. Well, nothing else is in the queue. So with that, I'll turn it back to Casey to thank our speakers and wrap up.

Thank you very much to Dr. Sabbagh and Mike Detke. I really appreciate everyone joining us today. Have a great one.

Thanks for including me.

Thanks.